colpali fix

pages/Multimodal_Conversational_Search.py

@@ -277,67 +277,22 @@ def render_answer(question,answer,index,res_img):
     with col2:
         ans_ = answer['answer']
         st.write(ans_)
-       
-        
-        
-        # def stream_():
-        #     #use for streaming response on the client side
-        #     for word in ans_.split(" "):
-        #         yield word + " "
-        #         time.sleep(0.04)
-        #     #use for streaming response from Llm directly
-        #     if(isinstance(ans_,botocore.eventstream.EventStream)):
-        #         for event in ans_:
-        #             chunk = event.get('chunk')
-                    
-        #             if chunk:
-                        
-        #                 chunk_obj = json.loads(chunk.get('bytes').decode())
-                        
-        #                 if('content_block' in chunk_obj or ('delta' in chunk_obj and 'text' in chunk_obj['delta'])):
-        #                     key_ = list(chunk_obj.keys())[2]
-        #                     text = chunk_obj[key_]['text']
-                            
-        #                     clear_output(wait=True)
-        #                     output.append(text)
-        #                     yield text
-        #                     time.sleep(0.04)
-            
-                
-        
-        # if(index == len(st.session_state.questions_)):
-        #     st.write_stream(stream_)
-        #     if(isinstance(st.session_state.answers_[index-1]['answer'],botocore.eventstream.EventStream)):
-        #         st.session_state.answers_[index-1]['answer'] = "".join(output)
-        # else:
-        #     st.write(ans_)
-        
-
-        # polly_response = polly_client.synthesize_speech(VoiceId='Joanna',
-        #                 OutputFormat='ogg_vorbis', 
-        #                 Text = ans_,
-        #                 Engine = 'neural')
-
-        # audio_col1, audio_col2 = st.columns([50,50])
-        # with audio_col1:
-        #     st.audio(polly_response['AudioStream'].read(), format="audio/ogg")
-            
         rdn_key_1 = ''.join([random.choice(string.ascii_letters)
                               for _ in range(10)])
         def show_maxsim():
             st.session_state.show_columns = True
             st.session_state.maxSimImages = colpali.img_highlight(st.session_state.top_img, st.session_state.query_token_vectors, st.session_state.query_tokens)
+            st.session_state.input_query = st.session_state.questions_[-1]["question"]
+            st.session_state.answers_.pop()
+            st.session_state.questions_.pop()
             handle_input()
             with placeholder.container():
-                render_all()    
+                render_all()
+              
         if(st.session_state.input_is_colpali):
             st.button("Show similarity map",key=rdn_key_1,on_click = show_maxsim)
                 
-        
-        
-        #st.markdown("<div style='font-size:18px;padding:3px 7px 3px 7px;borderWidth: 0px;borderColor: red;borderStyle: solid;border-radius: 10px;'>"+ans_+"</div>", unsafe_allow_html = True)
-    #st.markdown("<div style='color:#e28743';padding:3px 7px 3px 7px;borderWidth: 0px;borderColor: red;borderStyle: solid;width: fit-content;height: fit-content;border-radius: 10px;'><b>Relevant images from the document :</b></div>", unsafe_allow_html = True)
-    #st.write("")
+
     colu1,colu2,colu3 = st.columns([4,82,20])
     with colu2:
         with st.expander("Relevant Sources:"):
@@ -396,11 +351,6 @@ def render_answer(question,answer,index,res_img):
         
         
     with col_3:
-        
-        #st.markdown("<div style='color:#e28743;borderWidth: 0px;borderColor: red;borderStyle: solid;width: fit-content;height: fit-content;border-radius: 5px;'><b>"+",".join(st.session_state.input_rag_searchType)+"</b></div>", unsafe_allow_html = True)
-        
-       
-        
         if(index == len(st.session_state.questions_)):
 
             rdn_key = ''.join([random.choice(string.ascii_letters)